A New Algorithm Inspired on Reversible Elementary Cellular Automata for Global Optimization

This work presents a new global optimization algorithm of functions inspired by the dynamic behavior of reversible cellular automata, denominated Reversible Elementary Cellular Automata Algorithm (RECAA). This algorithm adapts the reversible evolution rules in elementary cellular automata (in one dimension and only with two states) to work with vectors of real values to realize optimization tasks. The originality of RECAA lies in adapting the dynamic of the reversible elementary cellular automata to perform exploration and exploitation actions in the optimization process. This work shows that diversity in cellular automata behaviors (in this case, reversibility) is useful to define new metaheuristics to solve optimization problems. The algorithm is compared with 15 recently published metaheuristics that recognized for their good performance, using 50 test functions in 30, 500, and with a fixed number of dimensions, and the CEC 2022 benchmark suit. Additionally, it is shown that RECAA has been applied in 3 engineering problems. In all the experiments, RECAA obtained satisfactory results. RECAA was implemented in MATLAB, and its source code can be consulted in GitHub.
https://github.com/juanseck/RECAA